INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

THE drosophila melanogaster Malpighian tubule is, on a per cell basis, the fastest-secreting epithelium known (12). Accordingly, it provides an excellent model for the analysis of epithelial function, with the added possibility of genetic analysis of the transport and control systems (9, 10, 14). As in nearly all insect (31) and many vertebrate (16, 18) epithelia, the very high rate of fluid secretion is energized by an apical plasma membrane vacuolar (V)-ATPase, driving a K⁺-H⁺ exchanger to achieve a net secretion of K⁺, entraining a flux of chloride and thus of water (9, 14, 21). Fluid secretion is under multiple hormonal control and is regulated by the second messengers adenosine 3',5'-cyclic monophosphate (cAMP), guanosine 3',5'-cyclic monophosphate (cGMP; derived from an endogenous nitric oxide synthase pathway), and intracellular calcium (7, 8, 13, 14, 21, 25).

In insect tubules, active transport of cations was assigned historically to the epithelial cells, whereas chloride, water, and solutes were thought to move paracellularly (23, 29). More recently, transcellular transports/channels have been discovered for several classes of molecules, offering an alternative route. For example, Drosophila tubules contain a water channel of the major intrinsic protein family (12) and a neuropeptide transporter (26); and a range of organic solutes are transported through the principal cells (27). It is thus of particular interest to determine the route of chloride transport through this epithelium, as flux rates are very high.

The chloride conductance of tubules of several insect orders is under hormonal control by members of the leucokinin (LK) neuropeptide family, which elicit a marked increase in fluid secretion rate, concomitant with a spectacular fall in the lumen-positive transepithelial potential toward zero (21). This collapse in potential is not reflected by intracellular potential measurements, which are stable throughout the course of LK stimulation; this has led most workers to hypothesize that the route of chloride flux must be paracellular (23). Additionally, several attempts to identify a chloride channel by patch clamping have been unsuccessful. However, in Drosophila tubules, the collapse in potential is complete within ~2 s after addition of LK (21), leading us to argue that this change is too rapid to be caused by an extensive remodelling of cell junctions and to propose that chloride flux is controlled by the small stellate cells (10, 21). Consistent with this, stellate cells are found only in the secretory region of the tubule (27).

Here, further evidence is presented in support of this hypothesis. Patch-clamp analysis shows that the tubule apical membrane contains a small relative area highly enriched in "maxi"-chloride channels, a distribution consistent with their localization to stellate cells. Supporting this result, both fluid secretion and transepithelial potential are exquisitely sensitive to classical chloride channel blockers, making a paracellular route less likely. Vibrating probe analysis of current densities around the basolateral surface of the tubule reveals spatial segregation of cation and anion transport, as has been hypothesized, with anion transport confined to stellate cells. Finally, using a novel transgenic aequorin system for calcium measurement we show that the effect of LK is to elevate calcium rapidly and selectively in stellate cells, consistent with a major role of these cells in regulating chloride conductance.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Experimental insects. Drosophila melanogaster (Oregon R strain) were maintained in laboratory cultures under standard conditions (1). Lines transgenic for GAL4 P-elements driving tubule-specific expression were those described by Sözen et al. (27), and upstream activating sequence (UAS)-aequorin lines were those described by Rosay et al. (25).

Patch-clamp analysis. Tubules were dissected in Drosophila saline (1), which consisted of (in mmol/l) 135 NaCl, 20 KCl, 2 CaCl₂, 8.5 MgCl₂, 10.2 NaHCO₃, 4.3 NaH₂PO₄, 15 N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid (HEPES), and 20 glucose. The tubules were allowed to adhere to tissue culture plastic and were exposed briefly to 1 mg/ml collagenase (Boehringer). The apical surface was exposed by cutting the tubule open longitudinally with a broken patch electrode.

Single-channel recording. The cell-attached and inside-out configurations of the patch-clamp technique were used for single-channel studies. Patch pipettes (tip resistance 4-5 M) were prepared from borosilicate glass capillaries (Vitrex; Modulohm, Herlev, Denmark), pulled, and polished on a programmable puller (DMZ; Werner Zeitz, Augsburg, Germany). The patch pipettes were filled with an intracellular-like K⁺ solution. This "K_int" pCa 6 solution contained the following (concentrations in mmol/l): 117 KCl, 4.29 MgCl₂, 1.03 CaCl₂, 10 HEPES, and 5 EGTA; the solution was adjusted to a final pH of 7.2. All solutions were filtered through 0.2-µm Millipore cellulose disks. Current and voltage recordings from patch-clamp electrodes were amplified (Biologic RK 400 patch-clamp amplifer; Claix, France), passed through a 30-kHz filter, and stored on digital audio tape (Sony, Tokyo, Japan). The single-channel current records were displayed in real time using a digital oscilloscope (model 310; Nicolet, Madison, WI). Single channel current records for Figs. 1-4 were filtered at 5 kHz using a Bessel filter and were replayed on a chart recorder (model RS3200; Gould Instruments, Valley View, OH).

Analysis. Current and voltage signals were digitized using a 16-bit analog-to-digital (A/D) converter (CED 1401; Cambridge Electronic Design, Cambridge UK) after low pass (3 dB) filtering at 5 kHz (8 pole Bessel filter, model 902; Frequency Devices, Haverhill, MA) and were analyzed with an 80486-50 MHz DELL personal computer. Current-voltage (I-V) relations were run and analyzed using VGEN and PAT programs (J. Dempster, Strathclyde Electrophysiology Software version 7.0; University of Strathclyde). Single-channel current amplitude histograms, open probability (P_o), and open and closed dwell times were analyzed from data segments of 60-120 s duration using the PAT program.

Cation movement across the membrane from the exterior (pipette) to cytoplasmic side is defined as inward current and is shown as downward deflections in single-channel recordings. Channel activity is expressed as N · P_o, where N is the number of simultaneously open channels present in the patch pipette verified by the number of Gaussian peaks detected on current amplitude histograms and P_o denotes the mean of open probability for individual channels.

Noise analysis. Whole cell current fluctuations were amplified by the patch-clamp amplifier and split into low- and high- pass components using Bessel filters with cutoff frequencies at 1 kHz and 0.1 Hz, respectively, and were stored on digital audiotape. The current-noise signal was replayed through the CED1401 A/D converter, digitized at 5 KHz, and analyzed by a fast-Fourier transform using customized software. The variance of the fluctuations in current noise is presented as a function of the alternating current frequencies contained in the current noise (power density spectra). An average power spectrum was compiled from 200 record segments (each of 250 ms duration), and channel noise was recorded at pipette holding potentials (V_p) of +50 mV relative to the Nernst potential for chloride ions (E_Cl) and at V_p = E_Cl (0 mV in symmetrical Cl solutions). A Lorentzian curve was fitted to the power spectrum using a Levenberg-Marquadt iterative fitting algorithm. For a channel that fluctuates between distinct open and closed states, the power spectrum will contain a Lorentzian curve defined by the equation

(1)

where S_o is the low frequency asymptote or plateau value and f_c the corner or cutoff frequency at which the spectral power is S_o/2. The variance in channel current-noise (²) was calculated by integrating the power spectrum

(2)

and the single-channel conductance (g_s) was then determined from

(3)

where I_m is the mean direct current recorded in the power spectrum, and V_p is the pipette holding voltage.

Fluid secretion assay. All experiments were performed at room temperature (22-28°C). Malpighian tubules were isolated from adult female flies dissected under standard saline. In some experiments, bathing saline chloride concentration was reduced 10-fold by mixing one part of standard saline with nine parts of chloride-free saline, which consisted of (in mmol/l) 135 sodium isethionate, 20 K₂SO₄, 2 CaSO₄, 8.5 MgSO₄, 10.2 NaHCO₃, 4.3 NaH₂PO₄, 15 HEPES, and 20 glucose. The pH of all salines was adjusted to 7.0. Except where noted, tubules were transferred into a standard bathing medium consisting of equal parts of Schneider's Drosophila medium and standard saline. Methods for collecting secreted fluid are described elsewhere (14). cAMP, cGMP, LK-I, and thapsigargin were obtained from Sigma Chemical, and LK-IV and LK-VI were obtained from Peninsula Laboratories. In Drosophila tubules, LK-I, -IV, and -VI are equally potent stimulants of fluid secretion.

Microelectrode analysis. Transepithelial potential differences (TEPs) were measured by inserting microelectrodes filled with 3 mol/l KCl into the lumen of the Malpighian tubule main segment. Dissected tubules were placed in saline in petri dishes in which 100-µl drops of 125 mg/ml poly-L-lysine had previously been placed and allowed to air-dry. Tubules readily adhered to the bottom of these dishes and did not move when the microelectrode tip was advanced against the tubule wall. Microelectrodes were advanced at an oblique angle using a hydraulic micromanipulator (Narishige, Tokyo, Japan) until a sudden shift in potential indicated that the basolateral membrane of a principal cell had been impaled. (Stellate cells, which could be distinguished by their less granular appearance under phase-contrast optics, were too small to permit reliable impalements with microelectrodes.) TEPs were measured by advancing the microelectrode tip further until the apical membrane was impaled and the microelectrode tip was positioned in the tubule lumen. Electrical potentials were recorded on a chart recorder or by a computerized data acquisition system (Axotape; Axon Instruments, Burlingame, CA).

Vibrating probe measurement of current densities. Tubules were pinned to apertures in Sylgard-coated microscope slides and perfused with saline as described previously (21). The probe was positioned one probe ball diameter (~20 µm) away from the surface of the tubule at an equatorial position. The probe measures extracellular current density in two dimensions (X and Z, where the length of the tubule is parallel to the y-axis) and displays the resulting vectors as X and Y. The length of the vector corresponds to the current density in microampere per square centimeter.

Calcium determination by directed expression of aequorin transgenes. The P{GAL4} enhancer trap technology allows the targeted expression of transgenes in intact organisms (2). A panel of such lines that can direct expression to specific cell types in the adult Drosophila Malpighian tubule has been generated (27). These have been used to drive expression of a UAS-aequorin transgene, allowing the cell specificity of tubule response to the neuropeptide CAP_2b to be studied (25). Here, the same technology (summarized in Fig. 1) is applied to demonstrate that LKs act to raise intracellular calcium in the stellate cells, using the P{GAL4} lines c710 or c724 to drive stellate cell specific expression (27). Thirty tubules, from 4- to 14-day-old adult progeny of a c710 × UAS-aequorin cross, were dissected in standard saline and incubated in the dark in Schneider's medium (Sigma) containing 2.5 µM coelenterazine for 4 h. Luminescence was measured with a luminometer (model LB9507; Berthold Wallac), which permitted the timed addition of agonists with continuous recording in 100-ms resolution. At the end of the experiment, total luminescence was measured over a 20-s period after the addition of 100 mM CaCl₂ and 1% Triton X-100, allowing the normalization of the counts to intracellular Ca²⁺ concentration ([Ca²⁺]_i) values.

View larger version (27K): [in this window] [in a new window]   Fig. 1.   Summary of GAL4-directed aequorin expression system for cell-specific calcium measurement. See MATERIALS AND METHODS for details.

Calcium activities were calculated based on a method described previously (6), using a program written in Perl.

Statistics. Data are shown as mean values ± SE, where n is the number of experiments. Where appropriate, the significance of differences was tested using Student's t-test (2-tailed) and setting a critical level of P = 0.05.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Patch-clamp analysis. Active cell-attached membrane patches contained a homogenous population of high-conductance channels. Multiple channels were usually seen in each (2-µm-diameter) patch (Fig. 2). Channel activity (single-channel P_o × number of active channels) was voltage dependent and highest (>0.9) at the resting membrane potential (Fig. 3A), with a mean open time of 13 ± 2 ms and a slope conductance of 256 ± 15 pS (Fig. 3B). At potentials more extreme than ±50 mV, the P_o was significantly reduced to <0.5 (Fig. 3A). In excised patches, the single-channel current-voltage relationship displayed a reversal potential of 38 mV (Fig. 3B), which was very close to the calculated chloride equilibrium potential between the pipette and bathing solutions (E_Cl = 40 mV). Based on these properties, we provisionally assign this channel as an insect homologue of a class of vertebrate maxi-chloride channels (5, 24). Such maxi-chloride channels are also known to be [Ca²⁺]_i activated; this would agree well with our experimental observations that tubule chloride conductance is stimulated by calcium and not by cyclic nucleotides (21), together with the demonstration here (see below) that LK acts to raise [Ca²⁺]_i in stellate cells.

View larger version (12K): [in this window] [in a new window]   Fig. 2.   Typical cell-attached patch records from Drosophila tubule apical membrane. Typical recording from a cell-attached patch. A: long-term recording (5 s/trace). B: short-term recording showing the presence of at least 3 active channels in the patch, with evidence for possible conductance substates. Scale bars denote 10 pA (vertical) and 100 ms (horizontal).

View larger version (13K): [in this window] [in a new window]   Fig. 3.   Channel characteristics. A: graph of N · Po against membrane voltage, showing the inverted U-shaped activation curve previously described for maxi-chloride channels (5), where N is the no. of simultaneously open channels and Po is open probability. B: mean single- channel current-voltage relation of maxi-chloride channel activity from 8 excised inside-out patches exposed to standard solution in patch pipette and intracellular-like K+ ("Kint") solution in the bath. The single- channel conductance is 256 pS, and the reversal potential is 38 mV. C: mean spectra densities of membrane current obtained at pipette holding potential (Vp)  ECl = 50 mV (filled symbols) and Vp = ECl (open symbols), where ECl denotes the Nernst potential for chloride ions. The line through the upper spectrum represents a single Lorentzian component L(f) = So/[1 + (f/fc)2] with corner frequency (fc) = 144 Hz and low frequency asymptote (So) = 5.24 pA2/Hz.

Is this Cl channel open under physiological conditions? Estimates for potential profiles across insect tubules of several other species suggest that the apical membrane potential would be +50 to +100 mV, lumen positive (15, 22, 30). If the activation properties of the channel in excised patches faithfully reflect those in vivo, then the channel is normally held at a condition of intermediate opening probability.

The relatively low fraction of patches to reveal maxi-chloride channels (5% of 172 patches) and the high channel densities (n = 5) in patches that did show activity also imply that the task of chloride transport may be localized spatially within tubules, consistent with our model for functional specialization between principal and secondary cells (9, 10, 21).

Noise analysis. Whole cell current fluctuations were recorded in situ using a patch electrode filling solution similar to "K_int, pCa6" with the addition of 10 mmol/l NaCl and standard Drosophila saline solution in the bath. Under these conditions, E_Na = 68 mV, E_K = 44 mV, and E_Cl = 6 mV. The ability to record single whole cell current fluctuations in situ indicates that those cells containing chloride channels are electrically uncoupled from surrounding cells or that uncoupling arises as a result of whole cell recording.

The power density spectrum of membrane current fluctuations could be fitted with a single Lorentzian component with a corner frequency, f_c, of 144 ± 9 Hz and a low-frequency asymptote, S_o, of 5.24 ± 0.18 pA² · Hz¹ · cell¹ (V_p = 50 mV relative to E_Cl, n = 6, Fig. 3C). The lower limit single-channel conductance, g_Cl(1  P_o), was 130 ± 9 pS at V_p = 50 mV (n = 6). With P_o = 0.5, this result indicates that Cl channels of large unitary conductance (200-300 pS) are present in this cell population.

At V_p 50 mV positive relative to E_Cl, the ratio of variance to mean membrane current was 6.5 pA, which decreased to 0.038 pA at V_p = E_Cl. Taken together, these results provide evidence that the membrane conductance is caused by the activation of a homogeneous population of relatively large maxi-chloride channels. Other channels, if present, contribute little to the membrane current.

Inhibition by extracellular 4,4'-diisothiocyanostilbene-2,2'-disulfonic acid. The maxi-chloride channel could be blocked by application of the stilbene derivative 4,4'-diisothiocyanostilbene-2,2'-disulfonic acid (DIDS) to the bath side of an outside-out membrane patch (Fig. 4). The inhibition of channel opening by DIDS was poorly reversible.

View larger version (16K): [in this window] [in a new window]   Fig. 4.   Effect of 4,4'-diisothiocyanostilbene-2,2'-disulfonic acid (DIDS) on apical chloride conductance. Current-event histograms of control outside-out membrane patch (A) and 60 s after addition of DIDS (100 µM; B) to the standard bathing solution (Kint solution in patch pipette, Vp = 20 mV). C: typical recording showing effect of DIDS. D: open probability (Prob) data, before and after addition of DIDS.

These data do not of themselves exclude an additional, paracellular route for chloride flux and do not address the nature of the chloride movement across the basolateral membrane; it proved difficult to obtain good basolateral seals, presumably because of difficulties in obtaining complete digestion of the basement membrane. However, these results identify a defined conductance that is present in an apical plasma membrane and is thus a plausible candidate for a transcellular route for chloride flux.

Fluid secretion assay. LK stimulates fluid secretion in tubules and collapses apical potentials without affecting the basolateral potential of principal cells (21, 23), leading several groups to suggest that chloride does not move through these cells (23). Fluid secretion by both unstimulated (Figs. 5, A-E) and LK-stimulated (Fig. 5F) Drosophila tubules was found to be sensitive to the classical chloride channel inhibitors diphenylamine-2-carboxylate, 5-nitro-2-(3-phenylpropylamino)benzoic acid, anthracene-9-carboxylic acid, and niflumic acid at concentrations at or below the levels effective in blocking chloride channels of vertebrate epithelial cells. This argues strongly for a role of chloride channels in transepithelial chloride flux.

View larger version (32K): [in this window] [in a new window]   Fig. 5.   Chloride channel blockers inhibit fluid secretion by Drosophila Malpighian tubules. A: time course showing fluid secretion rates of control tubules () and those treated with 0.1 mmol/l 5-nitro-2-(3-phenylpropylamino)benzoic acid (NPPB; ) at the time indicated by the vertical line. B-E: secretion rates of unstimulated tubules were significantly reduced (paired samples) 20-40 min after addition of diphenylamine-2-carboxylate (DPC; B), NPPB (C), anthracene-9-carboxylic acid (9-AC; D), or niflumic acid (E); n = 4-13 tubules (average 8) for each drug treatment (open bars) and 2-12 (average 6) for control tubules (filled bars). F: secretion rates for unstimulated tubules (filled bars), after stimulation with leucokinin (LK)-I (open bars) and after treatment with 100 µmol/l DPC, 20 µmol/l NPPB, 1 mmol/l 9-AC, and 20 µmol/l niflumic acid (hatched bars). LK increased fluid secretion rates significantly (paired samples) in all cases (filled vs. open bars). Secretion rates after the addition of each drug were significantly reduced (paired samples; open vs. filled bars) in all cases; n = 8-13 tubules (average 10) for each drug and 4-7 control tubules (not shown) for each drug.

Microelectrode analysis. Thapsigargin, an agent that mobilizes Ca²⁺ from intracellular stores (20), stimulates fluid secretion by Drosophila tubules while collapsing transepithelial potential (21). Although it is now clear that thapsigargin acts on both principal and stellate cells (25), monitoring of TEP allows its effects on shunt conductance to be studied separately from its modest effects on electrogenic transport. Long-term microelectrode impalements of tubules confirmed that the [Ca²⁺]_i-stimulated chloride shunt conductance was sensitive to the same drugs that inhibited fluid secretion (Fig. 6).

View larger version (22K): [in this window] [in a new window]   Fig. 6.   Chloride shunt conductance is sensitive to the chloride channel blocker NPPB. A: chloride shunt conductance was activated by the application of the intracellular Ca2+ concentration ([Ca2+]i)mobilizing agent thapsigargin to 50 µM at 350 s, resulting in a precipitous collapse of the lumen-positive potential around 900 s. The collapse was due to a chloride conductance, as exposure to low-chloride saline (at 1,350-1,450 s) reversibly returned the transepithelial potential (TEP) to its previous value. However, upon application of NPPB to 0.1 M, the effects of chloride substitution (at ~2,400 s) were massively attenuated; this effect was irreversible (see 2,600 s, after removal of NPPB at 2,500 s). [Cl]o, extracellular chloride concentration. B: summary of the effects of NPPB and niflumic acid on chloride-dependent changes in TEP for thapsigargin-stimulated tubules. Height of each bar shows the mean change in TEP (+1 SE) during a 10-fold reduction in bathing saline Cl concentration for control tubules (filled bars) and those exposed to 100 µM NPPB (n = 6) or 200 µM niflumic acid (n = 6; open bars). For both drugs, the reduction is significant.

From the fluid secretion and electrical data, we can conclude that the [Ca²⁺]_i-mediated increase in fluid secretion is mediated by a selective increase in chloride conductance that can be blocked by conventional chloride channel blockers. This is consistent with the patch-clamp data presented earlier (Figs. 2-6) but would be hard to reconcile with a paracellular route.

Vibrating probe measurement of current densities. If all cells in an epithelium were equally competent to transport ions, then the electrical field around the epithelium would be uniform. However, if these properties were spatially segregated, the fields would be nonuniform. The vibrating microelectrode technique is a powerful technique for the mapping of tiny electrical fields. When such analysis was performed on Drosophila Malpighian tubules, it was immediately clear that the anion conductance was confined to a relatively small number of hot spots within the main segment of the tubule, the region responsible for fluid secretion (Fig. 7).

View larger version (15K): [in this window] [in a new window]   Fig. 7.   Current density profile along the entire length of a single anterior Malpighian tubule. Clear initial segment of the tubule is visible on left, and the junction with the ureter is visible on right. Base of each arrow indicates the center point of each measurement. The probe measures extracellular current density in two dimensions (Y and Z, where the length of the tubule is parallel to the x-axis) and displays the resulting vectors as Y and X, respectively. The length of the vector corresponds to the current density in µA/cm2. Hot spots are confined to the main (secretory) segment (27) of the tubule.

In principle, both a paracellular route and a transcellular route through a small subset of cells could explain this effect. However, such profiles are not consistent with a uniform increase in the paracellular conductance. Enhancer trap analysis has shown that the main segment of each tubule contains 77 ± 1 principal cells and 22 ± 1 stellate cells (27), so the relatively low number of hot spots is only consistent with the chloride conductance being confined to a small subset of cells, probably the stellate cells. In this case, not all stellate cells can be equal in unstimulated tubules as, typically, there are usually four or five hot spots detected in scans of the whole tubule.

Further evidence to support this allocation can be seen in higher-power views of the hot spots. In each case, a stellate cell is found at the center of the hot spot (Fig. 8).

View larger version (41K): [in this window] [in a new window]   Fig. 8.   Chloride conductance hot spots colocalize with stellate cells. A-C: higher-magnification views (the diameter of the tubule can be taken to be ~35 µm in A-C) of individual anion conductance hot spots. In A, two of the arms of a stellate cell extend between the two asterisks. In B and C, the center of a stellate cell is indicated by an asterisk.

The vibrating probe detects field strength, not specific ions, and so it is important to distinguish a luminally directed anion current from a notional abluminally directed cation current. The hot spots are abolished by reduction of bathing chloride (Fig. 9, A and B) and by the same chloride channel blockers that inhibit fluid secretion (Fig. 9, C-E), so we can be confident that the hot spots are the sites of the [Ca²⁺]_i-induced chloride flux.

View larger version (61K): [in this window] [in a new window]   Fig. 9.   Stellate cell hot spot currents are carried by chloride. A: current density is significantly and reversibly reduced by a 10-fold reduction in bathing saline chloride concentration. Current density was measured at the same site on each tubule, which was bathed in control saline (CS) or in a saline containing of the control level of chloride. A: typical record. B: mean values + SE (n = 7 tubules). C and D: current density is significantly reduced by 100 µmol/l niflumic acid. C: typical record showing current density before (0) and 1, 3, 5, and 10 min after addition of niflumic acid. D: mean values + SE (n = 5 tubules) before and 10 min after addition of niflumic acid. E: current density is significantly reduced by 100 µmol/l NPPB. Mean values + SE (n = 5 tubules).

When tubules are stimulated with LK-I or thapsigargin, the average current density measured at hot spots identified in unstimulated tubules increases only modestly (data not shown), but currents at sites showing outward current density of 30-40 µA/cm² increase 25% or more (Fig. 10). What this suggests is that stimulation leads to recruitment, so that more stellate cells participate, but that the current at the hot spots of unstimulated tubules is already at or near maximum.

View larger version (15K): [in this window] [in a new window]   View larger version (16K): [in this window] [in a new window]   Fig. 10.   LK-I and thapisgargin significantly increase outward current density associated with stellate cells. A: in the absence of any stimulation, there is no significant change in the current density over 30 min (n = 42; 7 sites on 6 tubules). B: current densities 3 and 10 min after addition of 100 µmol/l LK-1 increase signficantly (n = 37; 3-9 sites on 6 tubules). C: current densities 5, 15, and 30 min after addition of 10 µmol/l thapsigargin increase signficantly (n = 37; 3-9 sites on 6 tubules). * P < 0.05 compared with corresponding controls by Student's t-test.

Calcium determination by directed expression of aequorin transgenes. These data demonstrate a role for the stellate cell in controlling [Ca²⁺]_i-activated anion conductance in the Malpighian tubule. If LK acts through this pathway, then it must raise [Ca²⁺]_i in the stellate cells. It is difficult to demonstrate this directly, because the stellate cells are too small to impale without damage and because the tubules actively transport fluorescent Ca²⁺ indicators too fast to allow accurate spectrophotometric measurements of [Ca²⁺]_i. Accordingly, we have applied a targeted expression system for the measurement of [Ca²⁺]_i in genetically defined subpopulations of cells in intact tissues that relies on an aequorin transgene (Fig. 1). The results show that stellate cell calcium levels rise significantly after application of LK-IV (Fig. 11). These data thus provide the first direct measurement of [Ca²⁺]_i in tubule stellate cells and the first direct demonstration that LK acts through [Ca²⁺]_i in stellate cells. Additionally, we can refine the rapid time course of LK action on transepithelial potential (21); the LK-induced [Ca²⁺]_i signal reaches a peak exceeding 200 nM within 100 ms after addition. As the luminometer has a sampling interval of 100 ms, it is possible that the actual Ca²⁺ response is still faster. This time course is faster than the earliest observed physiological response, the collapse of transapical potential (21), consistent with a role for [Ca²⁺]_i as a second messenger.

View larger version (11K): [in this window] [in a new window]   Fig. 11.   LK-IV acts to raise [Ca2+]i in stellate cells. A: typical response after a mock injection of saline (first arrow) or of saline containing LK (second arrow). Maximal increase was observed within one time interval (i.e., 100 ms) of addition. B: average maximum [Ca2+]i levels in resting tubules, after mock injections of saline, or after injection of LK-IV to 1 µM final concentration. Left, absolute [Ca2+]i activities; right, increases over resting levels elicited by mock or LK injections.

By contrast, LK action on principal cells was inconsistent (data not shown). An increase in [Ca²⁺]_i was only occasionally seen. It is possible that LKs may convey some other, longer-term signal to the principal cells; however, the data confirm that only in stellate cells was there a consistent rise in [Ca²⁺]_i that was fast enough to precede the rapid electrical effect.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

These data help to resolve a number of long-standing questions in insect physiology. The major counterion, chloride, flows through chloride channels of high unit conductance and with conventional pharmacology through a specialized secondary, or stellate cell, under the control of the LK neurohormone family, signaling obligatorily via intracellular calcium.

Although these data effectively rule out a general paracellular route for chloride in this tissue, it remains technically possible to construct a model in which chloride flows uniquely through the cell junctions surrounding stellate cells, thus approximating our results. Stellate cells would thus retain their role as regulators of this conductance, and calcium would retain its role as intracellular second messenger within stellate cells; thus, hot spots would be confined to relatively few sites that coincided with stellate cells. The balance of probabilities, however, weighs against this model: we have demonstrated the existence of a classical chloride channel, that the conductance is blocked by known chloride channel blockers, that the response is extremely rapid, and that the vibrating probe data show unimodal hot spots centered on each stellate cell, rather than bimodal peaks aligned to their edges. Our working model, therefore, is that both chloride and water fluxes are transcellular in the tubule.

There remain intriguing further problems to address. 1) Although the effects were not consistent, LK stimulation occasionally elicited [Ca²⁺]_i peaks in the principal cells. What is the nature of this message? Does it prime some aspect of principal cell function that does not directly stimulate fluid secretion? 2) LK stimulation appeared subjectively to increase the number, rather than the size, of the chloride conductance hot spots. Are stellate cells recruited individually on an all-or-nothing basis? There is support for this notion from recent work in Aedes; LK stimulation produced oscillations in apical potential (28). Principal cells may also be more active than was previously supposed; after stimulation of D. melanogaster tubules by the neuropeptide CAP_2b, both TEP (7) and intracellular calcium (25) frequently show oscillatory behavior. 3) Our model for transcellular chloride flux can be extended convincingly to tubules of other Dipteran genera, but what of other orders of insects that lack stellate cells? Are the stellate cells a unique specialization of this higher order of insects? Or, as has been hypothesized, does the discovery of structural complexity in Drosophila presage the discovery of latent heterogeneity in all insects (10, 27)? 4) It has already been shown that electrogenic cation transport is confined to the principal cells of the Drosophila tubule, and here we have shown that anion transport is assigned to the minor cell type. However, in nearly all epithelia of both invertebrates and vertebrates in which V-ATPases are active, they are found on the plasma membranes of only a subset of cells, rather than ubiquitously. For example, V-ATPases are confined to the apical membrane of goblet cells in Lepidopteran midgut (11, 19), to intercalated cells in kidney collecting duct (4) and epididymus (3), and the mitochondria-rich cells of frog skin (17). Is this specialization of roles a general rule for epithelial function, particularly in those epithelia energized by V-ATPases?